If you read a file and your system has sufficient resources, the OS will try to cache that file (at least partially) in memory. So if you do a XLookup to a .dtf from Sesame, it is likely (though by no means certain), that subsequent XLookups will be from a partial memory "image" of the file in RAM. This is the reason why it is so important to "shutdown" modern OSs before powering off computers - in that the OS may not have had time to "checkpoint" or update a file held in RAM back to the filesystem on a sudden power off.

If you need to guarantee that the file is in memory, and the XLookup is to a .dtf, you will need to set up a "RAM drive", wherein the OS allocates a specific portion of memory to act as a lettered (in WinDos) hard drive. This technique was quite common in the days of slow hard drives, but is getting rare now-a-days.

Note (for reference): If you are doing a XLookup to a Sesame .db file, it is automatically held in memory for subsequent lookups.

That is why only one server (even if standalone) is allowed to access any one application at a time. The "current" application is the one being accessed by its server (in memory), not the application file sitting on the hard drive. Of course, if any changes are made by that server, the change is written to the file immediately.